includes/upload/UploadBase.php

   1 <?php
   2 /**
   3  * Base class for the backend of file upload.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Upload
  22  */
  23 use MediaWiki\Shell\Shell;
  24
  25 /**
  26  * @defgroup Upload Upload related
  27  */
  28
  29 /**
  30  * @ingroup Upload
  31  *
  32  * UploadBase and subclasses are the backend of MediaWiki's file uploads.
  33  * The frontends are formed by ApiUpload and SpecialUpload.
  34  *
  35  * @author Brion Vibber
  36  * @author Bryan Tong Minh
  37  * @author Michael Dale
  38  */
  39 abstract class UploadBase {
  40         /** @var string Local file system path to the file to upload (or a local copy) */
  41         protected $mTempPath;
  42         /** @var TempFSFile|null Wrapper to handle deleting the temp file */
  43         protected $tempFileObj;
  44         /** @var string|null */
  45         protected $mDesiredDestName;
  46         /** @var string|null */
  47         protected $mDestName;
  48         /** @var string|null */
  49         protected $mRemoveTempFile;
  50         /** @var string|null */
  51         protected $mSourceType;
  52         /** @var Title|bool */
  53         protected $mTitle = false;
  54         /** @var int */
  55         protected $mTitleError = 0;
  56         /** @var string|null */
  57         protected $mFilteredName;
  58         /** @var string|null */
  59         protected $mFinalExtension;
  60         /** @var LocalFile */
  61         protected $mLocalFile;
  62         /** @var UploadStashFile */
  63         protected $mStashFile;
  64         /** @var int|null */
  65         protected $mFileSize;
  66         /** @var array|null */
  67         protected $mFileProps;
  68         /** @var string[] */
  69         protected $mBlackListedExtensions;
  70         /** @var bool|null */
  71         protected $mJavaDetected;
  72         /** @var string|null */
  73         protected $mSVGNSError;
  74
  75         protected static $safeXmlEncodings = [
  76                 'UTF-8',
  77                 'ISO-8859-1',
  78                 'ISO-8859-2',
  79                 'UTF-16',
  80                 'UTF-32',
  81                 'WINDOWS-1250',
  82                 'WINDOWS-1251',
  83                 'WINDOWS-1252',
  84                 'WINDOWS-1253',
  85                 'WINDOWS-1254',
  86                 'WINDOWS-1255',
  87                 'WINDOWS-1256',
  88                 'WINDOWS-1257',
  89                 'WINDOWS-1258',
  90         ];
  91
  92         const SUCCESS = 0;
  93         const OK = 0;
  94         const EMPTY_FILE = 3;
  95         const MIN_LENGTH_PARTNAME = 4;
  96         const ILLEGAL_FILENAME = 5;
  97         const OVERWRITE_EXISTING_FILE = 7; # Not used anymore; handled by verifyTitlePermissions()
  98         const FILETYPE_MISSING = 8;
  99         const FILETYPE_BADTYPE = 9;
 100         const VERIFICATION_ERROR = 10;
 101         const HOOK_ABORTED = 11;
 102         const FILE_TOO_LARGE = 12;
 103         const WINDOWS_NONASCII_FILENAME = 13;
 104         const FILENAME_TOO_LONG = 14;
 105
 106         /**
 107          * @param int $error
 108          * @return string
 109          */
 110         public function getVerificationErrorCode( $error ) {
 111                 $code_to_status = [
 112                         self::EMPTY_FILE => 'empty-file',
 113                         self::FILE_TOO_LARGE => 'file-too-large',
 114                         self::FILETYPE_MISSING => 'filetype-missing',
 115                         self::FILETYPE_BADTYPE => 'filetype-banned',
 116                         self::MIN_LENGTH_PARTNAME => 'filename-tooshort',
 117                         self::ILLEGAL_FILENAME => 'illegal-filename',
 118                         self::OVERWRITE_EXISTING_FILE => 'overwrite',
 119                         self::VERIFICATION_ERROR => 'verification-error',
 120                         self::HOOK_ABORTED => 'hookaborted',
 121                         self::WINDOWS_NONASCII_FILENAME => 'windows-nonascii-filename',
 122                         self::FILENAME_TOO_LONG => 'filename-toolong',
 123                 ];
 124                 return $code_to_status[$error] ?? 'unknown-error';
 125         }
 126
 127         /**
 128          * Returns true if uploads are enabled.
 129          * Can be override by subclasses.
 130          * @return bool
 131          */
 132         public static function isEnabled() {
 133                 global $wgEnableUploads;
 134
 135                 if ( !$wgEnableUploads ) {
 136                         return false;
 137                 }
 138
 139                 # Check php's file_uploads setting
 140                 return wfIsHHVM() || wfIniGetBool( 'file_uploads' );
 141         }
 142
 143         /**
 144          * Returns true if the user can use this upload module or else a string
 145          * identifying the missing permission.
 146          * Can be overridden by subclasses.
 147          *
 148          * @param User $user
 149          * @return bool|string
 150          */
 151         public static function isAllowed( $user ) {
 152                 foreach ( [ 'upload', 'edit' ] as $permission ) {
 153                         if ( !$user->isAllowed( $permission ) ) {
 154                                 return $permission;
 155                         }
 156                 }
 157
 158                 return true;
 159         }
 160
 161         /**
 162          * Returns true if the user has surpassed the upload rate limit, false otherwise.
 163          *
 164          * @param User $user
 165          * @return bool
 166          */
 167         public static function isThrottled( $user ) {
 168                 return $user->pingLimiter( 'upload' );
 169         }
 170
 171         // Upload handlers. Should probably just be a global.
 172         private static $uploadHandlers = [ 'Stash', 'File', 'Url' ];
 173
 174         /**
 175          * Create a form of UploadBase depending on wpSourceType and initializes it
 176          *
 177          * @param WebRequest &$request
 178          * @param string|null $type
 179          * @return null|UploadBase
 180          */
 181         public static function createFromRequest( &$request, $type = null ) {
 182                 $type = $type ?: $request->getVal( 'wpSourceType', 'File' );
 183
 184                 if ( !$type ) {
 185                         return null;
 186                 }
 187
 188                 // Get the upload class
 189                 $type = ucfirst( $type );
 190
 191                 // Give hooks the chance to handle this request
 192                 $className = null;
 193                 Hooks::run( 'UploadCreateFromRequest', [ $type, &$className ] );
 194                 if ( is_null( $className ) ) {
 195                         $className = 'UploadFrom' . $type;
 196                         wfDebug( __METHOD__ . ": class name: $className\n" );
 197                         if ( !in_array( $type, self::$uploadHandlers ) ) {
 198                                 return null;
 199                         }
 200                 }
 201
 202                 // Check whether this upload class is enabled
 203                 if ( !call_user_func( [ $className, 'isEnabled' ] ) ) {
 204                         return null;
 205                 }
 206
 207                 // Check whether the request is valid
 208                 if ( !call_user_func( [ $className, 'isValidRequest' ], $request ) ) {
 209                         return null;
 210                 }
 211
 212                 /** @var UploadBase $handler */
 213                 $handler = new $className;
 214
 215                 $handler->initializeFromRequest( $request );
 216
 217                 return $handler;
 218         }
 219
 220         /**
 221          * Check whether a request if valid for this handler
 222          * @param WebRequest $request
 223          * @return bool
 224          */
 225         public static function isValidRequest( $request ) {
 226                 return false;
 227         }
 228
 229         public function __construct() {
 230         }
 231
 232         /**
 233          * Returns the upload type. Should be overridden by child classes
 234          *
 235          * @since 1.18
 236          * @return string
 237          */
 238         public function getSourceType() {
 239                 return null;
 240         }
 241
 242         /**
 243          * Initialize the path information
 244          * @param string $name The desired destination name
 245          * @param string $tempPath The temporary path
 246          * @param int $fileSize The file size
 247          * @param bool $removeTempFile (false) remove the temporary file?
 248          * @throws MWException
 249          */
 250         public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
 251                 $this->mDesiredDestName = $name;
 252                 if ( FileBackend::isStoragePath( $tempPath ) ) {
 253                         throw new MWException( __METHOD__ . " given storage path `$tempPath`." );
 254                 }
 255
 256                 $this->setTempFile( $tempPath, $fileSize );
 257                 $this->mRemoveTempFile = $removeTempFile;
 258         }
 259
 260         /**
 261          * Initialize from a WebRequest. Override this in a subclass.
 262          *
 263          * @param WebRequest &$request
 264          */
 265         abstract public function initializeFromRequest( &$request );
 266
 267         /**
 268          * @param string $tempPath File system path to temporary file containing the upload
 269          * @param int|null $fileSize
 270          */
 271         protected function setTempFile( $tempPath, $fileSize = null ) {
 272                 $this->mTempPath = $tempPath;
 273                 $this->mFileSize = $fileSize ?: null;
 274                 if ( strlen( $this->mTempPath ) && file_exists( $this->mTempPath ) ) {
 275                         $this->tempFileObj = new TempFSFile( $this->mTempPath );
 276                         if ( !$fileSize ) {
 277                                 $this->mFileSize = filesize( $this->mTempPath );
 278                         }
 279                 } else {
 280                         $this->tempFileObj = null;
 281                 }
 282         }
 283
 284         /**
 285          * Fetch the file. Usually a no-op
 286          * @return Status
 287          */
 288         public function fetchFile() {
 289                 return Status::newGood();
 290         }
 291
 292         /**
 293          * Return true if the file is empty
 294          * @return bool
 295          */
 296         public function isEmptyFile() {
 297                 return empty( $this->mFileSize );
 298         }
 299
 300         /**
 301          * Return the file size
 302          * @return int
 303          */
 304         public function getFileSize() {
 305                 return $this->mFileSize;
 306         }
 307
 308         /**
 309          * Get the base 36 SHA1 of the file
 310          * @return string
 311          */
 312         public function getTempFileSha1Base36() {
 313                 return FSFile::getSha1Base36FromPath( $this->mTempPath );
 314         }
 315
 316         /**
 317          * @param string $srcPath The source path
 318          * @return string|bool The real path if it was a virtual URL Returns false on failure
 319          */
 320         public function getRealPath( $srcPath ) {
 321                 $repo = RepoGroup::singleton()->getLocalRepo();
 322                 if ( FileRepo::isVirtualUrl( $srcPath ) ) {
 323                         /** @todo Just make uploads work with storage paths UploadFromStash
 324                          *  loads files via virtual URLs.
 325                          */
 326                         $tmpFile = $repo->getLocalCopy( $srcPath );
 327                         if ( $tmpFile ) {
 328                                 $tmpFile->bind( $this ); // keep alive with $this
 329                         }
 330                         $path = $tmpFile ? $tmpFile->getPath() : false;
 331                 } else {
 332                         $path = $srcPath;
 333                 }
 334
 335                 return $path;
 336         }
 337
 338         /**
 339          * Verify whether the upload is sane.
 340          * @return mixed Const self::OK or else an array with error information
 341          */
 342         public function verifyUpload() {
 343                 /**
 344                  * If there was no filename or a zero size given, give up quick.
 345                  */
 346                 if ( $this->isEmptyFile() ) {
 347                         return [ 'status' => self::EMPTY_FILE ];
 348                 }
 349
 350                 /**
 351                  * Honor $wgMaxUploadSize
 352                  */
 353                 $maxSize = self::getMaxUploadSize( $this->getSourceType() );
 354                 if ( $this->mFileSize > $maxSize ) {
 355                         return [
 356                                 'status' => self::FILE_TOO_LARGE,
 357                                 'max' => $maxSize,
 358                         ];
 359                 }
 360
 361                 /**
 362                  * Look at the contents of the file; if we can recognize the
 363                  * type but it's corrupt or data of the wrong type, we should
 364                  * probably not accept it.
 365                  */
 366                 $verification = $this->verifyFile();
 367                 if ( $verification !== true ) {
 368                         return [
 369                                 'status' => self::VERIFICATION_ERROR,
 370                                 'details' => $verification
 371                         ];
 372                 }
 373
 374                 /**
 375                  * Make sure this file can be created
 376                  */
 377                 $result = $this->validateName();
 378                 if ( $result !== true ) {
 379                         return $result;
 380                 }
 381
 382                 return [ 'status' => self::OK ];
 383         }
 384
 385         /**
 386          * Verify that the name is valid and, if necessary, that we can overwrite
 387          *
 388          * @return mixed True if valid, otherwise and array with 'status'
 389          * and other keys
 390          */
 391         public function validateName() {
 392                 $nt = $this->getTitle();
 393                 if ( is_null( $nt ) ) {
 394                         $result = [ 'status' => $this->mTitleError ];
 395                         if ( $this->mTitleError == self::ILLEGAL_FILENAME ) {
 396                                 $result['filtered'] = $this->mFilteredName;
 397                         }
 398                         if ( $this->mTitleError == self::FILETYPE_BADTYPE ) {
 399                                 $result['finalExt'] = $this->mFinalExtension;
 400                                 if ( count( $this->mBlackListedExtensions ) ) {
 401                                         $result['blacklistedExt'] = $this->mBlackListedExtensions;
 402                                 }
 403                         }
 404
 405                         return $result;
 406                 }
 407                 $this->mDestName = $this->getLocalFile()->getName();
 408
 409                 return true;
 410         }
 411
 412         /**
 413          * Verify the MIME type.
 414          *
 415          * @note Only checks that it is not an evil MIME. The "does it have
 416          *  correct extension given its MIME type?" check is in verifyFile.
 417          *  in `verifyFile()` that MIME type and file extension correlate.
 418          * @param string $mime Representing the MIME
 419          * @return mixed True if the file is verified, an array otherwise
 420          */
 421         protected function verifyMimeType( $mime ) {
 422                 global $wgVerifyMimeType, $wgVerifyMimeTypeIE;
 423                 if ( $wgVerifyMimeType ) {
 424                         wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
 425                         global $wgMimeTypeBlacklist;
 426                         if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
 427                                 return [ 'filetype-badmime', $mime ];
 428                         }
 429
 430                         if ( $wgVerifyMimeTypeIE ) {
 431                                 # Check what Internet Explorer would detect
 432                                 $fp = fopen( $this->mTempPath, 'rb' );
 433                                 $chunk = fread( $fp, 256 );
 434                                 fclose( $fp );
 435
 436                                 $magic = MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer();
 437                                 $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
 438                                 $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
 439                                 foreach ( $ieTypes as $ieType ) {
 440                                         if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
 441                                                 return [ 'filetype-bad-ie-mime', $ieType ];
 442                                         }
 443                                 }
 444                         }
 445                 }
 446
 447                 return true;
 448         }
 449
 450         /**
 451          * Verifies that it's ok to include the uploaded file
 452          *
 453          * @return mixed True of the file is verified, array otherwise.
 454          */
 455         protected function verifyFile() {
 456                 global $wgVerifyMimeType, $wgDisableUploadScriptChecks;
 457
 458                 $status = $this->verifyPartialFile();
 459                 if ( $status !== true ) {
 460                         return $status;
 461                 }
 462
 463                 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer() );
 464                 $this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 465                 $mime = $this->mFileProps['mime'];
 466
 467                 if ( $wgVerifyMimeType ) {
 468                         # XXX: Missing extension will be caught by validateName() via getTitle()
 469                         if ( $this->mFinalExtension != '' && !$this->verifyExtension( $mime, $this->mFinalExtension ) ) {
 470                                 return [ 'filetype-mime-mismatch', $this->mFinalExtension, $mime ];
 471                         }
 472                 }
 473
 474                 # check for htmlish code and javascript
 475                 if ( !$wgDisableUploadScriptChecks ) {
 476                         if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 477                                 $svgStatus = $this->detectScriptInSvg( $this->mTempPath, false );
 478                                 if ( $svgStatus !== false ) {
 479                                         return $svgStatus;
 480                                 }
 481                         }
 482                 }
 483
 484                 $handler = MediaHandler::getHandler( $mime );
 485                 if ( $handler ) {
 486                         $handlerStatus = $handler->verifyUpload( $this->mTempPath );
 487                         if ( !$handlerStatus->isOK() ) {
 488                                 $errors = $handlerStatus->getErrorsArray();
 489
 490                                 return reset( $errors );
 491                         }
 492                 }
 493
 494                 $error = true;
 495                 Hooks::run( 'UploadVerifyFile', [ $this, $mime, &$error ] );
 496                 if ( $error !== true ) {
 497                         if ( !is_array( $error ) ) {
 498                                 $error = [ $error ];
 499                         }
 500                         return $error;
 501                 }
 502
 503                 wfDebug( __METHOD__ . ": all clear; passing.\n" );
 504
 505                 return true;
 506         }
 507
 508         /**
 509          * A verification routine suitable for partial files
 510          *
 511          * Runs the blacklist checks, but not any checks that may
 512          * assume the entire file is present.
 513          *
 514          * @return mixed True for valid or array with error message key.
 515          */
 516         protected function verifyPartialFile() {
 517                 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
 518
 519                 # getTitle() sets some internal parameters like $this->mFinalExtension
 520                 $this->getTitle();
 521
 522                 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer() );
 523                 $this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 524
 525                 # check MIME type, if desired
 526                 $mime = $this->mFileProps['file-mime'];
 527                 $status = $this->verifyMimeType( $mime );
 528                 if ( $status !== true ) {
 529                         return $status;
 530                 }
 531
 532                 # check for htmlish code and javascript
 533                 if ( !$wgDisableUploadScriptChecks ) {
 534                         if ( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
 535                                 return [ 'uploadscripted' ];
 536                         }
 537                         if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 538                                 $svgStatus = $this->detectScriptInSvg( $this->mTempPath, true );
 539                                 if ( $svgStatus !== false ) {
 540                                         return $svgStatus;
 541                                 }
 542                         }
 543                 }
 544
 545                 # Check for Java applets, which if uploaded can bypass cross-site
 546                 # restrictions.
 547                 if ( !$wgAllowJavaUploads ) {
 548                         $this->mJavaDetected = false;
 549                         $zipStatus = ZipDirectoryReader::read( $this->mTempPath,
 550                                 [ $this, 'zipEntryCallback' ] );
 551                         if ( !$zipStatus->isOK() ) {
 552                                 $errors = $zipStatus->getErrorsArray();
 553                                 $error = reset( $errors );
 554                                 if ( $error[0] !== 'zip-wrong-format' ) {
 555                                         return $error;
 556                                 }
 557                         }
 558                         if ( $this->mJavaDetected ) {
 559                                 return [ 'uploadjava' ];
 560                         }
 561                 }
 562
 563                 # Scan the uploaded file for viruses
 564                 $virus = $this->detectVirus( $this->mTempPath );
 565                 if ( $virus ) {
 566                         return [ 'uploadvirus', $virus ];
 567                 }
 568
 569                 return true;
 570         }
 571
 572         /**
 573          * Callback for ZipDirectoryReader to detect Java class files.
 574          *
 575          * @param array $entry
 576          */
 577         public function zipEntryCallback( $entry ) {
 578                 $names = [ $entry['name'] ];
 579
 580                 // If there is a null character, cut off the name at it, because JDK's
 581                 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
 582                 // were constructed which had ".class\0" followed by a string chosen to
 583                 // make the hash collide with the truncated name, that file could be
 584                 // returned in response to a request for the .class file.
 585                 $nullPos = strpos( $entry['name'], "\000" );
 586                 if ( $nullPos !== false ) {
 587                         $names[] = substr( $entry['name'], 0, $nullPos );
 588                 }
 589
 590                 // If there is a trailing slash in the file name, we have to strip it,
 591                 // because that's what ZIP_GetEntry() does.
 592                 if ( preg_grep( '!\.class/?$!', $names ) ) {
 593                         $this->mJavaDetected = true;
 594                 }
 595         }
 596
 597         /**
 598          * Alias for verifyTitlePermissions. The function was originally
 599          * 'verifyPermissions', but that suggests it's checking the user, when it's
 600          * really checking the title + user combination.
 601          *
 602          * @param User $user User object to verify the permissions against
 603          * @return mixed An array as returned by getUserPermissionsErrors or true
 604          *   in case the user has proper permissions.
 605          */
 606         public function verifyPermissions( $user ) {
 607                 return $this->verifyTitlePermissions( $user );
 608         }
 609
 610         /**
 611          * Check whether the user can edit, upload and create the image. This
 612          * checks only against the current title; if it returns errors, it may
 613          * very well be that another title will not give errors. Therefore
 614          * isAllowed() should be called as well for generic is-user-blocked or
 615          * can-user-upload checking.
 616          *
 617          * @param User $user User object to verify the permissions against
 618          * @return mixed An array as returned by getUserPermissionsErrors or true
 619          *   in case the user has proper permissions.
 620          */
 621         public function verifyTitlePermissions( $user ) {
 622                 /**
 623                  * If the image is protected, non-sysop users won't be able
 624                  * to modify it by uploading a new revision.
 625                  */
 626                 $nt = $this->getTitle();
 627                 if ( is_null( $nt ) ) {
 628                         return true;
 629                 }
 630                 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
 631                 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
 632                 if ( !$nt->exists() ) {
 633                         $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
 634                 } else {
 635                         $permErrorsCreate = [];
 636                 }
 637                 if ( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
 638                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
 639                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
 640
 641                         return $permErrors;
 642                 }
 643
 644                 $overwriteError = $this->checkOverwrite( $user );
 645                 if ( $overwriteError !== true ) {
 646                         return [ $overwriteError ];
 647                 }
 648
 649                 return true;
 650         }
 651
 652         /**
 653          * Check for non fatal problems with the file.
 654          *
 655          * This should not assume that mTempPath is set.
 656          *
 657          * @return mixed[] Array of warnings
 658          */
 659         public function checkWarnings() {
 660                 $warnings = [];
 661
 662                 $localFile = $this->getLocalFile();
 663                 $localFile->load( File::READ_LATEST );
 664                 $filename = $localFile->getName();
 665                 $hash = $this->getTempFileSha1Base36();
 666
 667                 $badFileName = $this->checkBadFileName( $filename, $this->mDesiredDestName );
 668                 if ( $badFileName !== null ) {
 669                         $warnings['badfilename'] = $badFileName;
 670                 }
 671
 672                 $unwantedFileExtensionDetails = $this->checkUnwantedFileExtensions( $this->mFinalExtension );
 673                 if ( $unwantedFileExtensionDetails !== null ) {
 674                         $warnings['filetype-unwanted-type'] = $unwantedFileExtensionDetails;
 675                 }
 676
 677                 $fileSizeWarnings = $this->checkFileSize( $this->mFileSize );
 678                 if ( $fileSizeWarnings ) {
 679                         $warnings = array_merge( $warnings, $fileSizeWarnings );
 680                 }
 681
 682                 $localFileExistsWarnings = $this->checkLocalFileExists( $localFile, $hash );
 683                 if ( $localFileExistsWarnings ) {
 684                         $warnings = array_merge( $warnings, $localFileExistsWarnings );
 685                 }
 686
 687                 if ( $this->checkLocalFileWasDeleted( $localFile ) ) {
 688                         $warnings['was-deleted'] = $filename;
 689                 }
 690
 691                 // If a file with the same name exists locally then the local file has already been tested
 692                 // for duplication of content
 693                 $ignoreLocalDupes = isset( $warnings[ 'exists '] );
 694                 $dupes = $this->checkAgainstExistingDupes( $hash, $ignoreLocalDupes );
 695                 if ( $dupes ) {
 696                         $warnings['duplicate'] = $dupes;
 697                 }
 698
 699                 $archivedDupes = $this->checkAgainstArchiveDupes( $hash );
 700                 if ( $archivedDupes !== null ) {
 701                         $warnings['duplicate-archive'] = $archivedDupes;
 702                 }
 703
 704                 return $warnings;
 705         }
 706
 707         /**
 708          * Convert the warnings array returned by checkWarnings() to something that
 709          * can be serialized. File objects will be converted to an associative array
 710          * with the following keys:
 711          *
 712          *   - fileName: The name of the file
 713          *   - timestamp: The upload timestamp
 714          *
 715          * @param mixed[] $warnings
 716          * @return mixed[]
 717          */
 718         public static function makeWarningsSerializable( $warnings ) {
 719                 array_walk_recursive( $warnings, function ( &$param, $key ) {
 720                         if ( $param instanceof File ) {
 721                                 $param = [
 722                                         'fileName' => $param->getName(),
 723                                         'timestamp' => $param->getTimestamp()
 724                                 ];
 725                         } elseif ( is_object( $param ) ) {
 726                                 throw new InvalidArgumentException(
 727                                         'UploadBase::makeWarningsSerializable: ' .
 728                                         'Unexpected object of class ' . get_class( $param ) );
 729                         }
 730                 } );
 731                 return $warnings;
 732         }
 733
 734         /**
 735          * Check whether the resulting filename is different from the desired one,
 736          * but ignore things like ucfirst() and spaces/underscore things
 737          *
 738          * @param string $filename
 739          * @param string $desiredFileName
 740          *
 741          * @return string|null String that was determined to be bad or null if the filename is okay
 742          */
 743         private function checkBadFileName( $filename, $desiredFileName ) {
 744                 $comparableName = str_replace( ' ', '_', $desiredFileName );
 745                 $comparableName = Title::capitalize( $comparableName, NS_FILE );
 746
 747                 if ( $desiredFileName != $filename && $comparableName != $filename ) {
 748                         return $filename;
 749                 }
 750
 751                 return null;
 752         }
 753
 754         /**
 755          * @param string $fileExtension The file extension to check
 756          *
 757          * @return array|null array with the following keys:
 758          *                    0 => string The final extension being used
 759          *                    1 => string[] The extensions that are allowed
 760          *                    2 => int The number of extensions that are allowed.
 761          */
 762         private function checkUnwantedFileExtensions( $fileExtension ) {
 763                 global $wgCheckFileExtensions, $wgFileExtensions, $wgLang;
 764
 765                 if ( $wgCheckFileExtensions ) {
 766                         $extensions = array_unique( $wgFileExtensions );
 767                         if ( !$this->checkFileExtension( $fileExtension, $extensions ) ) {
 768                                 return [
 769                                         $fileExtension,
 770                                         $wgLang->commaList( $extensions ),
 771                                         count( $extensions )
 772                                 ];
 773                         }
 774                 }
 775
 776                 return null;
 777         }
 778
 779         /**
 780          * @param int $fileSize
 781          *
 782          * @return array warnings
 783          */
 784         private function checkFileSize( $fileSize ) {
 785                 global $wgUploadSizeWarning;
 786
 787                 $warnings = [];
 788
 789                 if ( $wgUploadSizeWarning && ( $fileSize > $wgUploadSizeWarning ) ) {
 790                         $warnings['large-file'] = [ $wgUploadSizeWarning, $fileSize ];
 791                 }
 792
 793                 if ( $fileSize == 0 ) {
 794                         $warnings['empty-file'] = true;
 795                 }
 796
 797                 return $warnings;
 798         }
 799
 800         /**
 801          * @param LocalFile $localFile
 802          * @param string $hash sha1 hash of the file to check
 803          *
 804          * @return array warnings
 805          */
 806         private function checkLocalFileExists( LocalFile $localFile, $hash ) {
 807                 $warnings = [];
 808
 809                 $exists = self::getExistsWarning( $localFile );
 810                 if ( $exists !== false ) {
 811                         $warnings['exists'] = $exists;
 812
 813                         // check if file is an exact duplicate of current file version
 814                         if ( $hash === $localFile->getSha1() ) {
 815                                 $warnings['no-change'] = $localFile;
 816                         }
 817
 818                         // check if file is an exact duplicate of older versions of this file
 819                         $history = $localFile->getHistory();
 820                         foreach ( $history as $oldFile ) {
 821                                 if ( $hash === $oldFile->getSha1() ) {
 822                                         $warnings['duplicate-version'][] = $oldFile;
 823                                 }
 824                         }
 825                 }
 826
 827                 return $warnings;
 828         }
 829
 830         private function checkLocalFileWasDeleted( LocalFile $localFile ) {
 831                 return $localFile->wasDeleted() && !$localFile->exists();
 832         }
 833
 834         /**
 835          * @param string $hash sha1 hash of the file to check
 836          * @param bool $ignoreLocalDupes True to ignore local duplicates
 837          *
 838          * @return File[] Duplicate files, if found.
 839          */
 840         private function checkAgainstExistingDupes( $hash, $ignoreLocalDupes ) {
 841                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 842                 $title = $this->getTitle();
 843                 foreach ( $dupes as $key => $dupe ) {
 844                         if (
 845                                 ( $dupe instanceof LocalFile ) &&
 846                                 $ignoreLocalDupes &&
 847                                 $title->equals( $dupe->getTitle() )
 848                         ) {
 849                                 unset( $dupes[$key] );
 850                         }
 851                 }
 852
 853                 return $dupes;
 854         }
 855
 856         /**
 857          * @param string $hash sha1 hash of the file to check
 858          *
 859          * @return string|null Name of the dupe or empty string if discovered (depending on visibility)
 860          *                     null if the check discovered no dupes.
 861          */
 862         private function checkAgainstArchiveDupes( $hash ) {
 863                 $archivedFile = new ArchivedFile( null, 0, '', $hash );
 864                 if ( $archivedFile->getID() > 0 ) {
 865                         if ( $archivedFile->userCan( File::DELETED_FILE ) ) {
 866                                 return $archivedFile->getName();
 867                         } else {
 868                                 return '';
 869                         }
 870                 }
 871
 872                 return null;
 873         }
 874
 875         /**
 876          * Really perform the upload. Stores the file in the local repo, watches
 877          * if necessary and runs the UploadComplete hook.
 878          *
 879          * @param string $comment
 880          * @param string $pageText
 881          * @param bool $watch Whether the file page should be added to user's watchlist.
 882          *   (This doesn't check $user's permissions.)
 883          * @param User $user
 884          * @param string[] $tags Change tags to add to the log entry and page revision.
 885          *   (This doesn't check $user's permissions.)
 886          * @return Status Indicating the whether the upload succeeded.
 887          */
 888         public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
 889                 $this->getLocalFile()->load( File::READ_LATEST );
 890                 $props = $this->mFileProps;
 891
 892                 $error = null;
 893                 Hooks::run( 'UploadVerifyUpload', [ $this, $user, $props, $comment, $pageText, &$error ] );
 894                 if ( $error ) {
 895                         if ( !is_array( $error ) ) {
 896                                 $error = [ $error ];
 897                         }
 898                         return Status::newFatal( ...$error );
 899                 }
 900
 901                 $status = $this->getLocalFile()->upload(
 902                         $this->mTempPath,
 903                         $comment,
 904                         $pageText,
 905                         File::DELETE_SOURCE,
 906                         $props,
 907                         false,
 908                         $user,
 909                         $tags
 910                 );
 911
 912                 if ( $status->isGood() ) {
 913                         if ( $watch ) {
 914                                 WatchAction::doWatch(
 915                                         $this->getLocalFile()->getTitle(),
 916                                         $user,
 917                                         User::IGNORE_USER_RIGHTS
 918                                 );
 919                         }
 920                         // Avoid PHP 7.1 warning of passing $this by reference
 921                         $uploadBase = $this;
 922                         Hooks::run( 'UploadComplete', [ &$uploadBase ] );
 923
 924                         $this->postProcessUpload();
 925                 }
 926
 927                 return $status;
 928         }
 929
 930         /**
 931          * Perform extra steps after a successful upload.
 932          *
 933          * @since  1.25
 934          */
 935         public function postProcessUpload() {
 936         }
 937
 938         /**
 939          * Returns the title of the file to be uploaded. Sets mTitleError in case
 940          * the name was illegal.
 941          *
 942          * @return Title|null The title of the file or null in case the name was illegal
 943          */
 944         public function getTitle() {
 945                 if ( $this->mTitle !== false ) {
 946                         return $this->mTitle;
 947                 }
 948                 if ( !is_string( $this->mDesiredDestName ) ) {
 949                         $this->mTitleError = self::ILLEGAL_FILENAME;
 950                         $this->mTitle = null;
 951
 952                         return $this->mTitle;
 953                 }
 954                 /* Assume that if a user specified File:Something.jpg, this is an error
 955                  * and that the namespace prefix needs to be stripped of.
 956                  */
 957                 $title = Title::newFromText( $this->mDesiredDestName );
 958                 if ( $title && $title->getNamespace() == NS_FILE ) {
 959                         $this->mFilteredName = $title->getDBkey();
 960                 } else {
 961                         $this->mFilteredName = $this->mDesiredDestName;
 962                 }
 963
 964                 # oi_archive_name is max 255 bytes, which include a timestamp and an
 965                 # exclamation mark, so restrict file name to 240 bytes.
 966                 if ( strlen( $this->mFilteredName ) > 240 ) {
 967                         $this->mTitleError = self::FILENAME_TOO_LONG;
 968                         $this->mTitle = null;
 969
 970                         return $this->mTitle;
 971                 }
 972
 973                 /**
 974                  * Chop off any directories in the given filename. Then
 975                  * filter out illegal characters, and try to make a legible name
 976                  * out of it. We'll strip some silently that Title would die on.
 977                  */
 978                 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mFilteredName );
 979                 /* Normalize to title form before we do any further processing */
 980                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 981                 if ( is_null( $nt ) ) {
 982                         $this->mTitleError = self::ILLEGAL_FILENAME;
 983                         $this->mTitle = null;
 984
 985                         return $this->mTitle;
 986                 }
 987                 $this->mFilteredName = $nt->getDBkey();
 988
 989                 /**
 990                  * We'll want to blacklist against *any* 'extension', and use
 991                  * only the final one for the whitelist.
 992                  */
 993                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 994
 995                 if ( $ext !== [] ) {
 996                         $this->mFinalExtension = trim( end( $ext ) );
 997                 } else {
 998                         $this->mFinalExtension = '';
 999
1000                         # No extension, try guessing one
1001                         $magic = MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer();
1002                         $mime = $magic->guessMimeType( $this->mTempPath );
1003                         if ( $mime !== 'unknown/unknown' ) {
1004                                 # Get a space separated list of extensions
1005                                 $extList = $magic->getExtensionsForType( $mime );
1006                                 if ( $extList ) {
1007                                         # Set the extension to the canonical extension
1008                                         $this->mFinalExtension = strtok( $extList, ' ' );
1009
1010                                         # Fix up the other variables
1011                                         $this->mFilteredName .= ".{$this->mFinalExtension}";
1012                                         $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
1013                                         $ext = [ $this->mFinalExtension ];
1014                                 }
1015                         }
1016                 }
1017
1018                 /* Don't allow users to override the blacklist (check file extension) */
1019                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
1020                 global $wgFileExtensions, $wgFileBlacklist;
1021
1022                 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
1023
1024                 if ( $this->mFinalExtension == '' ) {
1025                         $this->mTitleError = self::FILETYPE_MISSING;
1026                         $this->mTitle = null;
1027
1028                         return $this->mTitle;
1029                 } elseif ( $blackListedExtensions ||
1030                         ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
1031                                 !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) )
1032                 ) {
1033                         $this->mBlackListedExtensions = $blackListedExtensions;
1034                         $this->mTitleError = self::FILETYPE_BADTYPE;
1035                         $this->mTitle = null;
1036
1037                         return $this->mTitle;
1038                 }
1039
1040                 // Windows may be broken with special characters, see T3780
1041                 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
1042                         && !RepoGroup::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
1043                 ) {
1044                         $this->mTitleError = self::WINDOWS_NONASCII_FILENAME;
1045                         $this->mTitle = null;
1046
1047                         return $this->mTitle;
1048                 }
1049
1050                 # If there was more than one "extension", reassemble the base
1051                 # filename to prevent bogus complaints about length
1052                 if ( count( $ext ) > 1 ) {
1053                         $iterations = count( $ext ) - 1;
1054                         for ( $i = 0; $i < $iterations; $i++ ) {
1055                                 $partname .= '.' . $ext[$i];
1056                         }
1057                 }
1058
1059                 if ( strlen( $partname ) < 1 ) {
1060                         $this->mTitleError = self::MIN_LENGTH_PARTNAME;
1061                         $this->mTitle = null;
1062
1063                         return $this->mTitle;
1064                 }
1065
1066                 $this->mTitle = $nt;
1067
1068                 return $this->mTitle;
1069         }
1070
1071         /**
1072          * Return the local file and initializes if necessary.
1073          *
1074          * @return LocalFile|null
1075          */
1076         public function getLocalFile() {
1077                 if ( is_null( $this->mLocalFile ) ) {
1078                         $nt = $this->getTitle();
1079                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
1080                 }
1081
1082                 return $this->mLocalFile;
1083         }
1084
1085         /**
1086          * @return UploadStashFile|null
1087          */
1088         public function getStashFile() {
1089                 return $this->mStashFile;
1090         }
1091
1092         /**
1093          * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must
1094          * be called before calling this method (unless $isPartial is true).
1095          *
1096          * Upload stash exceptions are also caught and converted to an error status.
1097          *
1098          * @since 1.28
1099          * @param User $user
1100          * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file).
1101          * @return Status If successful, value is an UploadStashFile instance
1102          */
1103         public function tryStashFile( User $user, $isPartial = false ) {
1104                 if ( !$isPartial ) {
1105                         $error = $this->runUploadStashFileHook( $user );
1106                         if ( $error ) {
1107                                 return Status::newFatal( ...$error );
1108                         }
1109                 }
1110                 try {
1111                         $file = $this->doStashFile( $user );
1112                         return Status::newGood( $file );
1113                 } catch ( UploadStashException $e ) {
1114                         return Status::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() );
1115                 }
1116         }
1117
1118         /**
1119          * @param User $user
1120          * @return array|null Error message and parameters, null if there's no error
1121          */
1122         protected function runUploadStashFileHook( User $user ) {
1123                 $props = $this->mFileProps;
1124                 $error = null;
1125                 Hooks::run( 'UploadStashFile', [ $this, $user, $props, &$error ] );
1126                 if ( $error && !is_array( $error ) ) {
1127                         $error = [ $error ];
1128                 }
1129                 return $error;
1130         }
1131
1132         /**
1133          * If the user does not supply all necessary information in the first upload
1134          * form submission (either by accident or by design) then we may want to
1135          * stash the file temporarily, get more information, and publish the file
1136          * later.
1137          *
1138          * This method will stash a file in a temporary directory for later
1139          * processing, and save the necessary descriptive info into the database.
1140          * This method returns the file object, which also has a 'fileKey' property
1141          * which can be passed through a form or API request to find this stashed
1142          * file again.
1143          *
1144          * @deprecated since 1.28 Use tryStashFile() instead
1145          * @param User|null $user
1146          * @return UploadStashFile Stashed file
1147          * @throws UploadStashBadPathException
1148          * @throws UploadStashFileException
1149          * @throws UploadStashNotLoggedInException
1150          */
1151         public function stashFile( User $user = null ) {
1152                 wfDeprecated( __METHOD__, '1.28' );
1153
1154                 return $this->doStashFile( $user );
1155         }
1156
1157         /**
1158          * Implementation for stashFile() and tryStashFile().
1159          *
1160          * @param User|null $user
1161          * @return UploadStashFile Stashed file
1162          */
1163         protected function doStashFile( User $user = null ) {
1164                 $stash = RepoGroup::singleton()->getLocalRepo()->getUploadStash( $user );
1165                 $file = $stash->stashFile( $this->mTempPath, $this->getSourceType() );
1166                 $this->mStashFile = $file;
1167
1168                 return $file;
1169         }
1170
1171         /**
1172          * If we've modified the upload file we need to manually remove it
1173          * on exit to clean up.
1174          */
1175         public function cleanupTempFile() {
1176                 if ( $this->mRemoveTempFile && $this->tempFileObj ) {
1177                         // Delete when all relevant TempFSFile handles go out of scope
1178                         wfDebug( __METHOD__ . ": Marked temporary file '{$this->mTempPath}' for removal\n" );
1179                         $this->tempFileObj->autocollect();
1180                 }
1181         }
1182
1183         public function getTempPath() {
1184                 return $this->mTempPath;
1185         }
1186
1187         /**
1188          * Split a file into a base name and all dot-delimited 'extensions'
1189          * on the end. Some web server configurations will fall back to
1190          * earlier pseudo-'extensions' to determine type and execute
1191          * scripts, so the blacklist needs to check them all.
1192          *
1193          * @param string $filename
1194          * @return array [ string, string[] ]
1195          */
1196         public static function splitExtensions( $filename ) {
1197                 $bits = explode( '.', $filename );
1198                 $basename = array_shift( $bits );
1199
1200                 return [ $basename, $bits ];
1201         }
1202
1203         /**
1204          * Perform case-insensitive match against a list of file extensions.
1205          * Returns true if the extension is in the list.
1206          *
1207          * @param string $ext
1208          * @param array $list
1209          * @return bool
1210          */
1211         public static function checkFileExtension( $ext, $list ) {
1212                 return in_array( strtolower( $ext ), $list );
1213         }
1214
1215         /**
1216          * Perform case-insensitive match against a list of file extensions.
1217          * Returns an array of matching extensions.
1218          *
1219          * @param string[] $ext
1220          * @param string[] $list
1221          * @return string[]
1222          */
1223         public static function checkFileExtensionList( $ext, $list ) {
1224                 return array_intersect( array_map( 'strtolower', $ext ), $list );
1225         }
1226
1227         /**
1228          * Checks if the MIME type of the uploaded file matches the file extension.
1229          *
1230          * @param string $mime The MIME type of the uploaded file
1231          * @param string $extension The filename extension that the file is to be served with
1232          * @return bool
1233          */
1234         public static function verifyExtension( $mime, $extension ) {
1235                 $magic = MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer();
1236
1237                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' ) {
1238                         if ( !$magic->isRecognizableExtension( $extension ) ) {
1239                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
1240                                         "unrecognized extension '$extension', can't verify\n" );
1241
1242                                 return true;
1243                         } else {
1244                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; " .
1245                                         "recognized extension '$extension', so probably invalid file\n" );
1246
1247                                 return false;
1248                         }
1249                 }
1250
1251                 $match = $magic->isMatchingExtension( $extension, $mime );
1252
1253                 if ( $match === null ) {
1254                         if ( $magic->getTypesForExtension( $extension ) !== null ) {
1255                                 wfDebug( __METHOD__ . ": No extension known for $mime, but we know a mime for $extension\n" );
1256
1257                                 return false;
1258                         } else {
1259                                 wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
1260
1261                                 return true;
1262                         }
1263                 } elseif ( $match === true ) {
1264                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
1265
1266                         /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1267                         return true;
1268                 } else {
1269                         wfDebug( __METHOD__
1270                                 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1271
1272                         return false;
1273                 }
1274         }
1275
1276         /**
1277          * Heuristic for detecting files that *could* contain JavaScript instructions or
1278          * things that may look like HTML to a browser and are thus
1279          * potentially harmful. The present implementation will produce false
1280          * positives in some situations.
1281          *
1282          * @param string $file Pathname to the temporary upload file
1283          * @param string $mime The MIME type of the file
1284          * @param string $extension The extension of the file
1285          * @return bool True if the file contains something looking like embedded scripts
1286          */
1287         public static function detectScript( $file, $mime, $extension ) {
1288                 # ugly hack: for text files, always look at the entire file.
1289                 # For binary field, just check the first K.
1290
1291                 $isText = strpos( $mime, 'text/' ) === 0;
1292                 if ( $isText ) {
1293                         $chunk = file_get_contents( $file );
1294                 } else {
1295                         $fp = fopen( $file, 'rb' );
1296                         $chunk = fread( $fp, 1024 );
1297                         fclose( $fp );
1298                 }
1299
1300                 $chunk = strtolower( $chunk );
1301
1302                 if ( !$chunk ) {
1303                         return false;
1304                 }
1305
1306                 # decode from UTF-16 if needed (could be used for obfuscation).
1307                 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1308                         $enc = 'UTF-16BE';
1309                 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1310                         $enc = 'UTF-16LE';
1311                 } else {
1312                         $enc = null;
1313                 }
1314
1315                 if ( $enc ) {
1316                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1317                 }
1318
1319                 $chunk = trim( $chunk );
1320
1321                 /** @todo FIXME: Convert from UTF-16 if necessary! */
1322                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
1323
1324                 # check for HTML doctype
1325                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1326                         return true;
1327                 }
1328
1329                 // Some browsers will interpret obscure xml encodings as UTF-8, while
1330                 // PHP/expat will interpret the given encoding in the xml declaration (T49304)
1331                 if ( $extension == 'svg' || strpos( $mime, 'image/svg' ) === 0 ) {
1332                         if ( self::checkXMLEncodingMissmatch( $file ) ) {
1333                                 return true;
1334                         }
1335                 }
1336
1337                 // Quick check for HTML heuristics in old IE and Safari.
1338                 //
1339                 // The exact heuristics IE uses are checked separately via verifyMimeType(), so we
1340                 // don't need them all here as it can cause many false positives.
1341                 //
1342                 // Check for `<script` and such still to forbid script tags and embedded HTML in SVG:
1343                 $tags = [
1344                         '<body',
1345                         '<head',
1346                         '<html', # also in safari
1347                         '<script', # also in safari
1348                 ];
1349
1350                 foreach ( $tags as $tag ) {
1351                         if ( strpos( $chunk, $tag ) !== false ) {
1352                                 wfDebug( __METHOD__ . ": found something that may make it be mistaken for html: $tag\n" );
1353
1354                                 return true;
1355                         }
1356                 }
1357
1358                 /*
1359                  * look for JavaScript
1360                  */
1361
1362                 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1363                 $chunk = Sanitizer::decodeCharReferences( $chunk );
1364
1365                 # look for script-types
1366                 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1367                         wfDebug( __METHOD__ . ": found script types\n" );
1368
1369                         return true;
1370                 }
1371
1372                 # look for html-style script-urls
1373                 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1374                         wfDebug( __METHOD__ . ": found html-style script urls\n" );
1375
1376                         return true;
1377                 }
1378
1379                 # look for css-style script-urls
1380                 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1381                         wfDebug( __METHOD__ . ": found css-style script urls\n" );
1382
1383                         return true;
1384                 }
1385
1386                 wfDebug( __METHOD__ . ": no scripts found\n" );
1387
1388                 return false;
1389         }
1390
1391         /**
1392          * Check a whitelist of xml encodings that are known not to be interpreted differently
1393          * by the server's xml parser (expat) and some common browsers.
1394          *
1395          * @param string $file Pathname to the temporary upload file
1396          * @return bool True if the file contains an encoding that could be misinterpreted
1397          */
1398         public static function checkXMLEncodingMissmatch( $file ) {
1399                 global $wgSVGMetadataCutoff;
1400                 $contents = file_get_contents( $file, false, null, 0, $wgSVGMetadataCutoff );
1401                 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1402
1403                 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1404                         if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1405                                 && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1406                         ) {
1407                                 wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1408
1409                                 return true;
1410                         }
1411                 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1412                         // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1413                         // bytes. There shouldn't be a legitimate reason for this to happen.
1414                         wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1415
1416                         return true;
1417                 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1418                         // EBCDIC encoded XML
1419                         wfDebug( __METHOD__ . ": EBCDIC Encoded XML\n" );
1420
1421                         return true;
1422                 }
1423
1424                 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1425                 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1426                 $attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ];
1427                 foreach ( $attemptEncodings as $encoding ) {
1428                         Wikimedia\suppressWarnings();
1429                         $str = iconv( $encoding, 'UTF-8', $contents );
1430                         Wikimedia\restoreWarnings();
1431                         if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1432                                 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1433                                         && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1434                                 ) {
1435                                         wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1436
1437                                         return true;
1438                                 }
1439                         } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1440                                 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1441                                 // bytes. There shouldn't be a legitimate reason for this to happen.
1442                                 wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1443
1444                                 return true;
1445                         }
1446                 }
1447
1448                 return false;
1449         }
1450
1451         /**
1452          * @param string $filename
1453          * @param bool $partial
1454          * @return bool|array
1455          */
1456         protected function detectScriptInSvg( $filename, $partial ) {
1457                 $this->mSVGNSError = false;
1458                 $check = new XmlTypeCheck(
1459                         $filename,
1460                         [ $this, 'checkSvgScriptCallback' ],
1461                         true,
1462                         [
1463                                 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback',
1464                                 'external_dtd_handler' => 'UploadBase::checkSvgExternalDTD',
1465                         ]
1466                 );
1467                 if ( $check->wellFormed !== true ) {
1468                         // Invalid xml (T60553)
1469                         // But only when non-partial (T67724)
1470                         return $partial ? false : [ 'uploadinvalidxml' ];
1471                 } elseif ( $check->filterMatch ) {
1472                         if ( $this->mSVGNSError ) {
1473                                 return [ 'uploadscriptednamespace', $this->mSVGNSError ];
1474                         }
1475
1476                         return $check->filterMatchType;
1477                 }
1478
1479                 return false;
1480         }
1481
1482         /**
1483          * Callback to filter SVG Processing Instructions.
1484          * @param string $target Processing instruction name
1485          * @param string $data Processing instruction attribute and value
1486          * @return bool|array
1487          */
1488         public static function checkSvgPICallback( $target, $data ) {
1489                 // Don't allow external stylesheets (T59550)
1490                 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1491                         return [ 'upload-scripted-pi-callback' ];
1492                 }
1493
1494                 return false;
1495         }
1496
1497         /**
1498          * Verify that DTD urls referenced are only the standard dtds
1499          *
1500          * Browsers seem to ignore external dtds. However just to be on the
1501          * safe side, only allow dtds from the svg standard.
1502          *
1503          * @param string $type PUBLIC or SYSTEM
1504          * @param string $publicId The well-known public identifier for the dtd
1505          * @param string $systemId The url for the external dtd
1506          * @return bool|array
1507          */
1508         public static function checkSvgExternalDTD( $type, $publicId, $systemId ) {
1509                 // This doesn't include the XHTML+MathML+SVG doctype since we don't
1510                 // allow XHTML anyways.
1511                 $allowedDTDs = [
1512                         'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd',
1513                         'http://www.w3.org/TR/2001/REC-SVG-20010904/DTD/svg10.dtd',
1514                         'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-basic.dtd',
1515                         'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-tiny.dtd',
1516                         // https://phabricator.wikimedia.org/T168856
1517                         'http://www.w3.org/TR/2001/PR-SVG-20010719/DTD/svg10.dtd',
1518                 ];
1519                 if ( $type !== 'PUBLIC'
1520                         || !in_array( $systemId, $allowedDTDs )
1521                         || strpos( $publicId, "-//W3C//" ) !== 0
1522                 ) {
1523                         return [ 'upload-scripted-dtd' ];
1524                 }
1525                 return false;
1526         }
1527
1528         /**
1529          * @todo Replace this with a whitelist filter!
1530          * @param string $element
1531          * @param array $attribs
1532          * @param string|null $data
1533          * @return bool|array
1534          */
1535         public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1536                 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1537
1538                 // We specifically don't include:
1539                 // http://www.w3.org/1999/xhtml (T62771)
1540                 static $validNamespaces = [
1541                         '',
1542                         'adobe:ns:meta/',
1543                         'http://creativecommons.org/ns#',
1544                         'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1545                         'http://ns.adobe.com/adobeillustrator/10.0/',
1546                         'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1547                         'http://ns.adobe.com/extensibility/1.0/',
1548                         'http://ns.adobe.com/flows/1.0/',
1549                         'http://ns.adobe.com/illustrator/1.0/',
1550                         'http://ns.adobe.com/imagereplacement/1.0/',
1551                         'http://ns.adobe.com/pdf/1.3/',
1552                         'http://ns.adobe.com/photoshop/1.0/',
1553                         'http://ns.adobe.com/saveforweb/1.0/',
1554                         'http://ns.adobe.com/variables/1.0/',
1555                         'http://ns.adobe.com/xap/1.0/',
1556                         'http://ns.adobe.com/xap/1.0/g/',
1557                         'http://ns.adobe.com/xap/1.0/g/img/',
1558                         'http://ns.adobe.com/xap/1.0/mm/',
1559                         'http://ns.adobe.com/xap/1.0/rights/',
1560                         'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1561                         'http://ns.adobe.com/xap/1.0/stype/font#',
1562                         'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1563                         'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1564                         'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1565                         'http://ns.adobe.com/xap/1.0/t/pg/',
1566                         'http://purl.org/dc/elements/1.1/',
1567                         'http://purl.org/dc/elements/1.1',
1568                         'http://schemas.microsoft.com/visio/2003/svgextensions/',
1569                         'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1570                         'http://taptrix.com/inkpad/svg_extensions',
1571                         'http://web.resource.org/cc/',
1572                         'http://www.freesoftware.fsf.org/bkchem/cdml',
1573                         'http://www.inkscape.org/namespaces/inkscape',
1574                         'http://www.opengis.net/gml',
1575                         'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1576                         'http://www.w3.org/2000/svg',
1577                         'http://www.w3.org/tr/rec-rdf-syntax/',
1578                         'http://www.w3.org/2000/01/rdf-schema#',
1579                 ];
1580
1581                 // Inkscape mangles namespace definitions created by Adobe Illustrator.
1582                 // This is nasty but harmless. (T144827)
1583                 $isBuggyInkscape = preg_match( '/^&(#38;)*ns_[a-z_]+;$/', $namespace );
1584
1585                 if ( !( $isBuggyInkscape || in_array( $namespace, $validNamespaces ) ) ) {
1586                         wfDebug( __METHOD__ . ": Non-svg namespace '$namespace' in uploaded file.\n" );
1587                         /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1588                         $this->mSVGNSError = $namespace;
1589
1590                         return true;
1591                 }
1592
1593                 /*
1594                  * check for elements that can contain javascript
1595                  */
1596                 if ( $strippedElement == 'script' ) {
1597                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
1598
1599                         return [ 'uploaded-script-svg', $strippedElement ];
1600                 }
1601
1602                 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1603                 #  <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1604                 if ( $strippedElement == 'handler' ) {
1605                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1606
1607                         return [ 'uploaded-script-svg', $strippedElement ];
1608                 }
1609
1610                 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1611                 if ( $strippedElement == 'stylesheet' ) {
1612                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1613
1614                         return [ 'uploaded-script-svg', $strippedElement ];
1615                 }
1616
1617                 # Block iframes, in case they pass the namespace check
1618                 if ( $strippedElement == 'iframe' ) {
1619                         wfDebug( __METHOD__ . ": iframe in uploaded file.\n" );
1620
1621                         return [ 'uploaded-script-svg', $strippedElement ];
1622                 }
1623
1624                 # Check <style> css
1625                 if ( $strippedElement == 'style'
1626                         && self::checkCssFragment( Sanitizer::normalizeCss( $data ) )
1627                 ) {
1628                         wfDebug( __METHOD__ . ": hostile css in style element.\n" );
1629                         return [ 'uploaded-hostile-svg' ];
1630                 }
1631
1632                 foreach ( $attribs as $attrib => $value ) {
1633                         $stripped = $this->stripXmlNamespace( $attrib );
1634                         $value = strtolower( $value );
1635
1636                         if ( substr( $stripped, 0, 2 ) == 'on' ) {
1637                                 wfDebug( __METHOD__
1638                                         . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1639
1640                                 return [ 'uploaded-event-handler-on-svg', $attrib, $value ];
1641                         }
1642
1643                         # Do not allow relative links, or unsafe url schemas.
1644                         # For <a> tags, only data:, http: and https: and same-document
1645                         # fragment links are allowed. For all other tags, only data:
1646                         # and fragment are allowed.
1647                         if ( $stripped == 'href'
1648                                 && $value !== ''
1649                                 && strpos( $value, 'data:' ) !== 0
1650                                 && strpos( $value, '#' ) !== 0
1651                         ) {
1652                                 if ( !( $strippedElement === 'a'
1653                                         && preg_match( '!^https?://!i', $value ) )
1654                                 ) {
1655                                         wfDebug( __METHOD__ . ": Found href attribute <$strippedElement "
1656                                                 . "'$attrib'='$value' in uploaded file.\n" );
1657
1658                                         return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ];
1659                                 }
1660                         }
1661
1662                         # only allow data: targets that should be safe. This prevents vectors like,
1663                         # image/svg, text/xml, application/xml, and text/html, which can contain scripts
1664                         if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1665                                 // rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1666                                 // phpcs:ignore Generic.Files.LineLength
1667                                 $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1668
1669                                 if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) {
1670                                         wfDebug( __METHOD__ . ": Found href to unwhitelisted data: uri "
1671                                                 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1672                                         return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ];
1673                                 }
1674                         }
1675
1676                         # Change href with animate from (http://html5sec.org/#137).
1677                         if ( $stripped === 'attributename'
1678                                 && $strippedElement === 'animate'
1679                                 && $this->stripXmlNamespace( $value ) == 'href'
1680                         ) {
1681                                 wfDebug( __METHOD__ . ": Found animate that might be changing href using from "
1682                                         . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1683
1684                                 return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ];
1685                         }
1686
1687                         # use set/animate to add event-handler attribute to parent
1688                         if ( ( $strippedElement == 'set' || $strippedElement == 'animate' )
1689                                 && $stripped == 'attributename'
1690                                 && substr( $value, 0, 2 ) == 'on'
1691                         ) {
1692                                 wfDebug( __METHOD__ . ": Found svg setting event-handler attribute with "
1693                                         . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1694
1695                                 return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ];
1696                         }
1697
1698                         # use set to add href attribute to parent element
1699                         if ( $strippedElement == 'set'
1700                                 && $stripped == 'attributename'
1701                                 && strpos( $value, 'href' ) !== false
1702                         ) {
1703                                 wfDebug( __METHOD__ . ": Found svg setting href attribute '$value' in uploaded file.\n" );
1704
1705                                 return [ 'uploaded-setting-href-svg' ];
1706                         }
1707
1708                         # use set to add a remote / data / script target to an element
1709                         if ( $strippedElement == 'set'
1710                                 && $stripped == 'to'
1711                                 && preg_match( '!(http|https|data|script):!sim', $value )
1712                         ) {
1713                                 wfDebug( __METHOD__ . ": Found svg setting attribute to '$value' in uploaded file.\n" );
1714
1715                                 return [ 'uploaded-wrong-setting-svg', $value ];
1716                         }
1717
1718                         # use handler attribute with remote / data / script
1719                         if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1720                                 wfDebug( __METHOD__ . ": Found svg setting handler with remote/data/script "
1721                                         . "'$attrib'='$value' in uploaded file.\n" );
1722
1723                                 return [ 'uploaded-setting-handler-svg', $attrib, $value ];
1724                         }
1725
1726                         # use CSS styles to bring in remote code
1727                         if ( $stripped == 'style'
1728                                 && self::checkCssFragment( Sanitizer::normalizeCss( $value ) )
1729                         ) {
1730                                 wfDebug( __METHOD__ . ": Found svg setting a style with "
1731                                         . "remote url '$attrib'='$value' in uploaded file.\n" );
1732                                 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1733                         }
1734
1735                         # Several attributes can include css, css character escaping isn't allowed
1736                         $cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker',
1737                                 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' ];
1738                         if ( in_array( $stripped, $cssAttrs )
1739                                 && self::checkCssFragment( $value )
1740                         ) {
1741                                 wfDebug( __METHOD__ . ": Found svg setting a style with "
1742                                         . "remote url '$attrib'='$value' in uploaded file.\n" );
1743                                 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1744                         }
1745
1746                         # image filters can pull in url, which could be svg that executes scripts
1747                         # Only allow url( "#foo" ). Do not allow url( http://example.com )
1748                         if ( $strippedElement == 'image'
1749                                 && $stripped == 'filter'
1750                                 && preg_match( '!url\s*\(\s*["\']?[^#]!sim', $value )
1751                         ) {
1752                                 wfDebug( __METHOD__ . ": Found image filter with url: "
1753                                         . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1754
1755                                 return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1756                         }
1757                 }
1758
1759                 return false; // No scripts detected
1760         }
1761
1762         /**
1763          * Check a block of CSS or CSS fragment for anything that looks like
1764          * it is bringing in remote code.
1765          * @param string $value a string of CSS
1766          * @return bool true if the CSS contains an illegal string, false if otherwise
1767          */
1768         private static function checkCssFragment( $value ) {
1769                 # Forbid external stylesheets, for both reliability and to protect viewer's privacy
1770                 if ( stripos( $value, '@import' ) !== false ) {
1771                         return true;
1772                 }
1773
1774                 # We allow @font-face to embed fonts with data: urls, so we snip the string
1775                 # 'url' out so this case won't match when we check for urls below
1776                 $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1777                 $value = preg_replace( $pattern, '$1$2', $value );
1778
1779                 # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1780                 # properties filter and accelerator don't seem to be useful for xss in SVG files.
1781                 # Expression and -o-link don't seem to work either, but filtering them here in case.
1782                 # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1783                 # but not local ones such as url("#..., url('#..., url(#....
1784                 if ( preg_match( '!expression
1785                                 | -o-link\s*:
1786                                 | -o-link-source\s*:
1787                                 | -o-replace\s*:!imx', $value ) ) {
1788                         return true;
1789                 }
1790
1791                 if ( preg_match_all(
1792                                 "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1793                                 $value,
1794                                 $matches
1795                         ) !== 0
1796                 ) {
1797                         # TODO: redo this in one regex. Until then, url("#whatever") matches the first
1798                         foreach ( $matches[1] as $match ) {
1799                                 if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1800                                         return true;
1801                                 }
1802                         }
1803                 }
1804
1805                 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1806                         return true;
1807                 }
1808
1809                 return false;
1810         }
1811
1812         /**
1813          * Divide the element name passed by the xml parser to the callback into URI and prifix.
1814          * @param string $element
1815          * @return array Containing the namespace URI and prefix
1816          */
1817         private static function splitXmlNamespace( $element ) {
1818                 // 'http://www.w3.org/2000/svg:script' -> [ 'http://www.w3.org/2000/svg', 'script' ]
1819                 $parts = explode( ':', strtolower( $element ) );
1820                 $name = array_pop( $parts );
1821                 $ns = implode( ':', $parts );
1822
1823                 return [ $ns, $name ];
1824         }
1825
1826         /**
1827          * @param string $name
1828          * @return string
1829          */
1830         private function stripXmlNamespace( $name ) {
1831                 // 'http://www.w3.org/2000/svg:script' -> 'script'
1832                 $parts = explode( ':', strtolower( $name ) );
1833
1834                 return array_pop( $parts );
1835         }
1836
1837         /**
1838          * Generic wrapper function for a virus scanner program.
1839          * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1840          * $wgAntivirusRequired may be used to deny upload if the scan fails.
1841          *
1842          * @param string $file Pathname to the temporary upload file
1843          * @return bool|null|string False if not virus is found, null if the scan fails or is disabled,
1844          *   or a string containing feedback from the virus scanner if a virus was found.
1845          *   If textual feedback is missing but a virus was found, this function returns true.
1846          */
1847         public static function detectVirus( $file ) {
1848                 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1849
1850                 if ( !$wgAntivirus ) {
1851                         wfDebug( __METHOD__ . ": virus scanner disabled\n" );
1852
1853                         return null;
1854                 }
1855
1856                 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1857                         wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
1858                         $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1859                                 [ 'virus-badscanner', $wgAntivirus ] );
1860
1861                         return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1862                 }
1863
1864                 # look up scanner configuration
1865                 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1866                 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1867                 $msgPattern = $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ?? null;
1868
1869                 if ( strpos( $command, "%f" ) === false ) {
1870                         # simple pattern: append file to scan
1871                         $command .= " " . Shell::escape( $file );
1872                 } else {
1873                         # complex pattern: replace "%f" with file to scan
1874                         $command = str_replace( "%f", Shell::escape( $file ), $command );
1875                 }
1876
1877                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
1878
1879                 # execute virus scanner
1880                 $exitCode = false;
1881
1882                 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1883                 #      that does not seem to be worth the pain.
1884                 #      Ask me (Duesentrieb) about it if it's ever needed.
1885                 $output = wfShellExecWithStderr( $command, $exitCode );
1886
1887                 # map exit code to AV_xxx constants.
1888                 $mappedCode = $exitCode;
1889                 if ( $exitCodeMap ) {
1890                         if ( isset( $exitCodeMap[$exitCode] ) ) {
1891                                 $mappedCode = $exitCodeMap[$exitCode];
1892                         } elseif ( isset( $exitCodeMap["*"] ) ) {
1893                                 $mappedCode = $exitCodeMap["*"];
1894                         }
1895                 }
1896
1897                 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1898                  * so we need the strict equalities === and thus can't use a switch here
1899                  */
1900                 if ( $mappedCode === AV_SCAN_FAILED ) {
1901                         # scan failed (code was mapped to false by $exitCodeMap)
1902                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
1903
1904                         $output = $wgAntivirusRequired
1905                                 ? wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1906                                 : null;
1907                 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
1908                         # scan failed because filetype is unknown (probably imune)
1909                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
1910                         $output = null;
1911                 } elseif ( $mappedCode === AV_NO_VIRUS ) {
1912                         # no virus found
1913                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
1914                         $output = false;
1915                 } else {
1916                         $output = trim( $output );
1917
1918                         if ( !$output ) {
1919                                 $output = true; # if there's no output, return true
1920                         } elseif ( $msgPattern ) {
1921                                 $groups = [];
1922                                 if ( preg_match( $msgPattern, $output, $groups ) && $groups[1] ) {
1923                                         $output = $groups[1];
1924                                 }
1925                         }
1926
1927                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
1928                 }
1929
1930                 return $output;
1931         }
1932
1933         /**
1934          * Check if there's an overwrite conflict and, if so, if restrictions
1935          * forbid this user from performing the upload.
1936          *
1937          * @param User $user
1938          *
1939          * @return bool|array
1940          */
1941         private function checkOverwrite( $user ) {
1942                 // First check whether the local file can be overwritten
1943                 $file = $this->getLocalFile();
1944                 $file->load( File::READ_LATEST );
1945                 if ( $file->exists() ) {
1946                         if ( !self::userCanReUpload( $user, $file ) ) {
1947                                 return [ 'fileexists-forbidden', $file->getName() ];
1948                         } else {
1949                                 return true;
1950                         }
1951                 }
1952
1953                 /* Check shared conflicts: if the local file does not exist, but
1954                  * wfFindFile finds a file, it exists in a shared repository.
1955                  */
1956                 $file = wfFindFile( $this->getTitle(), [ 'latest' => true ] );
1957                 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1958                         return [ 'fileexists-shared-forbidden', $file->getName() ];
1959                 }
1960
1961                 return true;
1962         }
1963
1964         /**
1965          * Check if a user is the last uploader
1966          *
1967          * @param User $user
1968          * @param File $img
1969          * @return bool
1970          */
1971         public static function userCanReUpload( User $user, File $img ) {
1972                 if ( $user->isAllowed( 'reupload' ) ) {
1973                         return true; // non-conditional
1974                 } elseif ( !$user->isAllowed( 'reupload-own' ) ) {
1975                         return false;
1976                 }
1977
1978                 if ( !( $img instanceof LocalFile ) ) {
1979                         return false;
1980                 }
1981
1982                 $img->load();
1983
1984                 return $user->getId() == $img->getUser( 'id' );
1985         }
1986
1987         /**
1988          * Helper function that does various existence checks for a file.
1989          * The following checks are performed:
1990          * - The file exists
1991          * - Article with the same name as the file exists
1992          * - File exists with normalized extension
1993          * - The file looks like a thumbnail and the original exists
1994          *
1995          * @param File $file The File object to check
1996          * @return mixed False if the file does not exists, else an array
1997          */
1998         public static function getExistsWarning( $file ) {
1999                 if ( $file->exists() ) {
2000                         return [ 'warning' => 'exists', 'file' => $file ];
2001                 }
2002
2003                 if ( $file->getTitle()->getArticleID() ) {
2004                         return [ 'warning' => 'page-exists', 'file' => $file ];
2005                 }
2006
2007                 if ( strpos( $file->getName(), '.' ) == false ) {
2008                         $partname = $file->getName();
2009                         $extension = '';
2010                 } else {
2011                         $n = strrpos( $file->getName(), '.' );
2012                         $extension = substr( $file->getName(), $n + 1 );
2013                         $partname = substr( $file->getName(), 0, $n );
2014                 }
2015                 $normalizedExtension = File::normalizeExtension( $extension );
2016
2017                 if ( $normalizedExtension != $extension ) {
2018                         // We're not using the normalized form of the extension.
2019                         // Normal form is lowercase, using most common of alternate
2020                         // extensions (eg 'jpg' rather than 'JPEG').
2021
2022                         // Check for another file using the normalized form...
2023                         $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
2024                         $file_lc = wfLocalFile( $nt_lc );
2025
2026                         if ( $file_lc->exists() ) {
2027                                 return [
2028                                         'warning' => 'exists-normalized',
2029                                         'file' => $file,
2030                                         'normalizedFile' => $file_lc
2031                                 ];
2032                         }
2033                 }
2034
2035                 // Check for files with the same name but a different extension
2036                 $similarFiles = RepoGroup::singleton()->getLocalRepo()->findFilesByPrefix(
2037                         "{$partname}.", 1 );
2038                 if ( count( $similarFiles ) ) {
2039                         return [
2040                                 'warning' => 'exists-normalized',
2041                                 'file' => $file,
2042                                 'normalizedFile' => $similarFiles[0],
2043                         ];
2044                 }
2045
2046                 if ( self::isThumbName( $file->getName() ) ) {
2047                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
2048                         $nt_thb = Title::newFromText(
2049                                 substr( $partname, strpos( $partname, '-' ) + 1 ) . '.' . $extension,
2050                                 NS_FILE
2051                         );
2052                         $file_thb = wfLocalFile( $nt_thb );
2053                         if ( $file_thb->exists() ) {
2054                                 return [
2055                                         'warning' => 'thumb',
2056                                         'file' => $file,
2057                                         'thumbFile' => $file_thb
2058                                 ];
2059                         } else {
2060                                 // File does not exist, but we just don't like the name
2061                                 return [
2062                                         'warning' => 'thumb-name',
2063                                         'file' => $file,
2064                                         'thumbFile' => $file_thb
2065                                 ];
2066                         }
2067                 }
2068
2069                 foreach ( self::getFilenamePrefixBlacklist() as $prefix ) {
2070                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
2071                                 return [
2072                                         'warning' => 'bad-prefix',
2073                                         'file' => $file,
2074                                         'prefix' => $prefix
2075                                 ];
2076                         }
2077                 }
2078
2079                 return false;
2080         }
2081
2082         /**
2083          * Helper function that checks whether the filename looks like a thumbnail
2084          * @param string $filename
2085          * @return bool
2086          */
2087         public static function isThumbName( $filename ) {
2088                 $n = strrpos( $filename, '.' );
2089                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
2090
2091                 return (
2092                                 substr( $partname, 3, 3 ) == 'px-' ||
2093                                 substr( $partname, 2, 3 ) == 'px-'
2094                         ) &&
2095                         preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
2096         }
2097
2098         /**
2099          * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
2100          *
2101          * @return array List of prefixes
2102          */
2103         public static function getFilenamePrefixBlacklist() {
2104                 $blacklist = [];
2105                 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
2106                 if ( !$message->isDisabled() ) {
2107                         $lines = explode( "\n", $message->plain() );
2108                         foreach ( $lines as $line ) {
2109                                 // Remove comment lines
2110                                 $comment = substr( trim( $line ), 0, 1 );
2111                                 if ( $comment == '#' || $comment == '' ) {
2112                                         continue;
2113                                 }
2114                                 // Remove additional comments after a prefix
2115                                 $comment = strpos( $line, '#' );
2116                                 if ( $comment > 0 ) {
2117                                         $line = substr( $line, 0, $comment - 1 );
2118                                 }
2119                                 $blacklist[] = trim( $line );
2120                         }
2121                 }
2122
2123                 return $blacklist;
2124         }
2125
2126         /**
2127          * Gets image info about the file just uploaded.
2128          *
2129          * Also has the effect of setting metadata to be an 'indexed tag name' in
2130          * returned API result if 'metadata' was requested. Oddly, we have to pass
2131          * the "result" object down just so it can do that with the appropriate
2132          * format, presumably.
2133          *
2134          * @param ApiResult $result
2135          * @return array Image info
2136          */
2137         public function getImageInfo( $result ) {
2138                 $localFile = $this->getLocalFile();
2139                 $stashFile = $this->getStashFile();
2140                 // Calling a different API module depending on whether the file was stashed is less than optimal.
2141                 // In fact, calling API modules here at all is less than optimal. Maybe it should be refactored.
2142                 if ( $stashFile ) {
2143                         $imParam = ApiQueryStashImageInfo::getPropertyNames();
2144                         $info = ApiQueryStashImageInfo::getInfo( $stashFile, array_flip( $imParam ), $result );
2145                 } else {
2146                         $imParam = ApiQueryImageInfo::getPropertyNames();
2147                         $info = ApiQueryImageInfo::getInfo( $localFile, array_flip( $imParam ), $result );
2148                 }
2149
2150                 return $info;
2151         }
2152
2153         /**
2154          * @param array $error
2155          * @return Status
2156          */
2157         public function convertVerifyErrorToStatus( $error ) {
2158                 $code = $error['status'];
2159                 unset( $code['status'] );
2160
2161                 return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
2162         }
2163
2164         /**
2165          * Get the MediaWiki maximum uploaded file size for given type of upload, based on
2166          * $wgMaxUploadSize.
2167          *
2168          * @param null|string $forType
2169          * @return int
2170          */
2171         public static function getMaxUploadSize( $forType = null ) {
2172                 global $wgMaxUploadSize;
2173
2174                 if ( is_array( $wgMaxUploadSize ) ) {
2175                         if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
2176                                 return $wgMaxUploadSize[$forType];
2177                         } else {
2178                                 return $wgMaxUploadSize['*'];
2179                         }
2180                 } else {
2181                         return intval( $wgMaxUploadSize );
2182                 }
2183         }
2184
2185         /**
2186          * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
2187          * limit can't be guessed, returns a very large number (PHP_INT_MAX).
2188          *
2189          * @since 1.27
2190          * @return int
2191          */
2192         public static function getMaxPhpUploadSize() {
2193                 $phpMaxFileSize = wfShorthandToInteger(
2194                         ini_get( 'upload_max_filesize' ) ?: ini_get( 'hhvm.server.upload.upload_max_file_size' ),
2195                         PHP_INT_MAX
2196                 );
2197                 $phpMaxPostSize = wfShorthandToInteger(
2198                         ini_get( 'post_max_size' ) ?: ini_get( 'hhvm.server.max_post_size' ),
2199                         PHP_INT_MAX
2200                 ) ?: PHP_INT_MAX;
2201                 return min( $phpMaxFileSize, $phpMaxPostSize );
2202         }
2203
2204         /**
2205          * Get the current status of a chunked upload (used for polling)
2206          *
2207          * The value will be read from cache.
2208          *
2209          * @param User $user
2210          * @param string $statusKey
2211          * @return Status[]|bool
2212          */
2213         public static function getSessionStatus( User $user, $statusKey ) {
2214                 $store = self::getUploadSessionStore();
2215                 $key = self::getUploadSessionKey( $store, $user, $statusKey );
2216
2217                 return $store->get( $key );
2218         }
2219
2220         /**
2221          * Set the current status of a chunked upload (used for polling)
2222          *
2223          * The value will be set in cache for 1 day
2224          *
2225          * Avoid triggering this method on HTTP GET/HEAD requests
2226          *
2227          * @param User $user
2228          * @param string $statusKey
2229          * @param array|bool $value
2230          * @return void
2231          */
2232         public static function setSessionStatus( User $user, $statusKey, $value ) {
2233                 $store = self::getUploadSessionStore();
2234                 $key = self::getUploadSessionKey( $store, $user, $statusKey );
2235
2236                 if ( $value === false ) {
2237                         $store->delete( $key );
2238                 } else {
2239                         $store->set( $key, $value, $store::TTL_DAY );
2240                 }
2241         }
2242
2243         /**
2244          * @param BagOStuff $store
2245          * @param User $user
2246          * @param string $statusKey
2247          * @return string
2248          */
2249         private static function getUploadSessionKey( BagOStuff $store, User $user, $statusKey ) {
2250                 return $store->makeKey(
2251                         'uploadstatus',
2252                         $user->getId() ?: md5( $user->getName() ),
2253                         $statusKey
2254                 );
2255         }
2256
2257         /**
2258          * @return BagOStuff
2259          */
2260         private static function getUploadSessionStore() {
2261                 return ObjectCache::getInstance( 'db-replicated' );
2262         }
2263 }